Use of AI

Effective from 1 January 2026 · Next review: 1 January 2027

This policy sets out how authors, reviewers and editors of the journal “Indo-Iranian Languages” may use generative artificial intelligence (AI) tools, in particular large language models and chatbots. The policy is based on the COPE position statement “Authorship and AI tools” (2023), the WAME recommendations on chatbots and generative AI, and the STM “Recommendations for a Classification of AI Use in Academic Manuscript Preparation” (2025). Its key principles are human accountability, transparency, confidentiality and the integrity of linguistic data. Responsible use of AI in line with the rules below is permitted and does not affect the editorial assessment of a manuscript.

For authors

  • Artificial intelligence systems (ChatGPT, Claude, DeepSeek, etc.) cannot be authors or co-authors of an article, as they cannot take responsibility for the work.
  • AI may be used without disclosure: to check spelling, punctuation and grammar, to format the reference list and bring the manuscript into line with the journal’s requirements, and for technical formatting of data (e.g. transliteration, glossing).
  • AI may be used with a brief disclosure: to translate text written by the author, for stylistic editing, and to search the literature.
  • Computational tools used to collect, process or analyse data (e.g. automatic speech recognition, automatic segmentation and alignment of audio, transcription, automatic annotation or glossing, the use of language models as an object or instrument of research) must be described in the methods section, stating the tool, its version, the settings or prompts used, and how the results were verified.
  • AI may not be used: to create the substantive scholarly content of the manuscript (argumentation, analysis, interpretation or conclusions); to create, alter or supplement primary linguistic data; for references to literature that have not been checked against the original sources; to write reviews or assess scholarly work; or for any action that entails human responsibility.
  • The author bears full responsibility for the accuracy of the text, language examples, glossing and references, as well as for all passages prepared with the help of AI and for any related breaches of publication ethics.
  • Non-existent sources or data generated by AI are treated as falsification.
  • The use of AI is disclosed in a separate section placed immediately before the reference list. The disclosure states the name and version of the tool and the purpose and extent of its use.

Recommended wording:

  • In preparing this work, the author(s) used [name and version of the tool] for [purpose, e.g. language editing of the English text / translation of the manuscript from Russian]. The author(s) reviewed and edited the output and take full responsibility for the content of the publication.

 

For reviewers and editors

  • The editorial board may use AI-generated text detection tools; however, their results must be assessed by people rather than applied automatically, since such tools may wrongly flag texts by authors writing in a non-native language. Any doubts are discussed with the authors before a decision is made.
  • If undisclosed use of AI is suspected, the editorial board may ask the author for an explanation and the source materials.
  • Manuscripts and reviews must not be uploaded, in whole or in part, to publicly available AI services, as this breaches confidentiality.
  • Manuscript assessment and editorial decisions are made by people.

 

Breaches of this policy

Alleged breaches are handled in accordance with the journal’s publication ethics policy and COPE guidelines. Depending on the circumstances, they may result in: a request to add a missing disclosure; publication of a correction; rejection of the manuscript; or retraction of a published article if the use of AI has called into question the reliability of the data, analysis or references. Authors are always given the opportunity to explain.

This policy is reviewed at least once a year and updated to reflect developments in tools and professional guidance. The version in force on the date of submission applies.